Promoting effect of water in ruthenium-catalyzed hydrogenation of carbon dioxide to formic acid

Article abstract of DOI:10.1021/om000944x

A strong promoting effect of water in the catalytic hydrogenation of CO₂ to formic acid with the solvento metal hydride species TpRu(PPh₃)(CH₃CN)H is observed. High-pressure NMR monitoring of the catalytic reaction shows that CO₂ readily inserts into Ru-H to form the metal formate TpRu(PPh₃)(CH₃CN)(η¹-OCHO)· H₂O, in which the formate ligand is intermolecularly hydrogen-bonded to a water molecule. Theoretical calculations carried out at the B3LYP level show that reaction barrier of the CO₂ insertion is significantly reduced in the presence of water. In the transition state of the process, electrophilicity of the carbon center of CO₂ is enhanced by the formation of hydrogen bonds between its oxygen atoms and H₂O. The metal formato species comes into equilibrium with another metal formate rapidly; the second formato species TpRu(PPh₃)(H₂O)(η¹-OCHO) contains a coordinated H₂O, which is intramolecularly hydrogen-bonded with the formate ligand. In view of the stability of these two metal formates under catalytic conditions, it is very likely that they are not within the major catalytic cycle of the reaction. A catalytic cycle, which accounts for the promoting effect of water, is proposed. The key species in the cycle is the aquo metal hydride species TpRu(PPh₃)(H₂O)H, which could be generated by a ligand displacement reaction of TpRu(PPh₃)(CH₃CN)H with H₂O. It is proposed that TpRu(PPh₃)(H₂O)H is able to transfer a proton and a hydride simultaneously to CO₂ to yield formic acid in a concerted manner, itself being converted to a transient hydroxo species, which then associates a H₂ molecule. The aquo hydride complex TpRu(PPh₃)(H₂O) is regenerated via σ-metathesis between the hydroxo and η²-H₂ ligands. Theoretical calculations have been carried out to study the structural and energetic aspects of species involved in this catalytic cycle.

Full text of DOI:10.1021/om000944x

1216
Organometallics 2001, 20, 1216-1222
P r om otin g Effect of Wa ter in Ru th en iu m -Ca ta lyzed
Hyd r ogen a tion of Ca r bon Dioxid e to F or m ic Acid
Chuanqi Yin,† Zhitao Xu,‡ Sheng-Yong Yang,‡ Siu Man Ng,† Kwok Yin Wong,†
Zhenyang Lin,*,‡ and Chak Po Lau*,†
Department of Applied Biology & Chemical Technology, The Hong Kong Polytechnic
University, Hung Hom, Kowloon, Hong Kong, China, and Department of Chemistry,
The Hong Kong University of Science & Technology, Clear Water Bay,
Kowloon, Hong Kong, China
Received November 6, 2000
A strong promoting effect of water in the catalytic hydrogenation of CO2 to formic acid
with the solvento metal hydride species TpRu(PPh3)(CH3CN)H is observed. High-pressure
NMR monitoring of the catalytic reaction shows that CO2 readily inserts into Ru-H to form
the metal formate TpRu(PPh3)(CH3CN)(η1-OCHO)‚H2O, in which the formate ligand is
intermolecularly hydrogen-bonded to a water molecule. Theoretical calculations carried out
at the B3LYP level show that reaction barrier of the CO2 insertion is significantly reduced
in the presence of water. In the transition state of the process, electrophilicity of the carbon
center of CO2 is enhanced by the formation of hydrogen bonds between its oxygen atoms
and H2O. The metal formato species comes into equilibrium with another metal formate
rapidly; the second formato species TpRu(PPh3)(H2O)(η1-OCHO) contains a coordinated H2O,
which is intramolecularly hydrogen-bonded with the formate ligand. In view of the stability
of these two metal formates under catalytic conditions, it is very likely that they are not
within the major catalytic cycle of the reaction. A catalytic cycle, which accounts for the
promoting effect of water, is proposed. The key species in the cycle is the aquo metal hydride
species TpRu(PPh3)(H2O)H, which could be generated by a ligand displacement reaction of
TpRu(PPh3)(CH3CN)H with H2O. It is proposed that TpRu(PPh3)(H2O)H is able to transfer
a proton and a hydride simultaneously to CO2 to yield formic acid in a concerted manner,
itself being converted to a transient hydroxo species, which then associates a H2 molecule.
The aquo hydride complex TpRu(PPh3)(H2O)H is regenerated via σ-metathesis between the
hydroxo and η2-H2 ligands. Theoretical calculations have been carried out to study the
structural and energetic aspects of species involved in this catalytic cycle.
In tr od u ction
been provided to explain the positive “water effect”, it
has been speculated that hydrogen-bonding interaction
between an H2O molecule and an oxygen atom of CO2
enhances the electrophilicity at carbon and facilitates
its insertion into the metal-hydride bond.3
Transition-metal-catalyzed CO2 hydrogenation to for-
mic acid is one of the interesting and attractive reactions
of CO2 fixation. It has attracted much attention in
recent years, as it provides a promising approach for
utilizing carbon dioxide as a raw material for large-scale
chemical synthesis.1 It was shown in one of the early
papers that addition of a small quantity of water was
effective to enhance the catalytic hydrogenation of CO2
to formic acid by combinations of group VIII transition
metals and bases.2 The accelerating effect of small
amounts of added water in organic solvents has also
been observed in active Rh3 and Ru4 systems. Although
no compelling experimental and theoretical data have
We have recently found that the ruthenium hydride
complex TpRu(PPh3)(CH3CN)H (1; Tp ) hydrotris-
(pyrazolyl)borate) is a versatile complex exhibiting a
wide range of reactivity.5 It gives, upon protonation with
a strong acid such as HBF4‚Et2O or CF3SO3H, the η2-
dihydrogen complex [TpRu(PPh3)(CH3CN)(H2)]+, which
has been demonstrated to play an important role in the
catalytic hydrogenation of olefins.5a It is able to de-
carbonylate primary alcohols, RCH2OH, forming the
carbonyl complexes TpRu(PPh3)(CO)R.5b It also reacts
with silane, R3SiH, to give the highly fluxional TpRu-
(PPh3)“H2SiR3” complex, which contains a η2-silane
moiety.5c Continuing our investigation of the reactivity
of 1, we studied the catalytic activity of this complex in
† The Hong Kong Polytechnic University.
‡ The Hong Kong University of Science & Technology.
(1) (a) Behr, A. Carbon Dioxide Activation by Metal Complexes;
VCH: Weinheim, Germany, 1988. (b) Aresta, M., Schloss, J . V., Eds.
Enzymatic and Model Carboxylation and Reduction Reactions for
Carbon Dioxide Utilization; NATO ASI Series C, No. 314; Kluwer
Academic: Dordrecht, The Netherlands, 1990. (c) J essop, P. G.; Ikariya,
T.; Noyori, R. Chem. Rev. 1995, 95, 259. (d) Leitner, W. Angew. Chem.,
Int. Ed. Engl. 1995, 34, 2207. (e) Leitner, W. Coord. Chem. Rev. 1996,
153, 257.
(4) J essop, P. G.; Hsiao, Y.; Ikariya, T.; Noyori, R. J . Am. Chem.
Soc. 1996, 118, 344.
(5) (a) Chan, W. C.; Lau, C. P.; Chen, Y. Z.; Fan, Y. Q.; Ng, S. M.;
J ia, G. Organometallics 1997, 16, 34. (b) Chen, Y. Z.; Chan, W. C.;
Lau, C. P.; Chu, H. S.; Lee, H. L.; J ia, G. Organometallics 1997, 16,
1241. (c) Ng, S. M.; Lau, C. P.; Fan, M.-F.; Lin, Z. Organometallics
1999, 18, 2484.
(2) Inoue, Y.; Izumida, H.; Sasaki, Y.; Hashimoto, H. Chem. Lett.
1976, 863.
(3) Tsai, J .-C.; Nicholas, K. M. J . Am. Chem. Soc. 1992, 114, 5117.
10.1021/om000944x CCC: $20.00 © 2001 American Chemical Society
Publication on Web 02/16/2001

Ru-Catalyzed Hydrogenation of Carbon Dioxide
Organometallics, Vol. 20, No. 6, 2001 1217
Ta ble 1. Hyd r ogen a tion of CO2 to F or m ic Acid
w ith Tp Ru (P P h 3)(CH3CN)H (1)a
of Et3N, was pressurized with CO2/H2 (8 atm/8 atm) at
room temperature in a 5 mm Wilmad pressure-valved
NMR tube. The solution was allowed to stand at room
temperature for 30 min, after which 31P{1H} NMR
spectroscopy showed that 1 remained unchanged. How-
ever, the 31P{1H} NMR spectrum taken at 80 °C after
the solution was heated at this temperature for 10 min
showed that most of the complex 1 was converted to a
mixture of TpRu(PPh3)(H2)H (2; δ 75.0 ppm) and
another species 3 corresponding to a signal at δ 54.9
ppm. Complex 2 was formed by displacement of CH3CN
with H2; its presence was further confirmed by observa-
tion of its hydride signal at δ -9.80 ppm (d, J (HP) )
entry
reacn time (h)
turnover no.b
1c,d
2d
3e
4f
5
16
16
16
16
4
trace
30
720
735
250
436
760
346
146
6
8
7
16
16
16
8g
9h
a
Typical reaction conditions: catalyst, 0.013 mmol; NEt3, 2 mL;
solvent, THF/H2O (15 mL/5 mL); total pressure, 50 atm (CO2/H2
b
) 1/1); temperature, 100 °C. TON ) mol of product/mol of 1.c No
1
18.4 Hz) in H NMR spectroscopy. We have previously
d
NEt3 added. Solvent THF (20 mL). e Solvent THF/H2O (19 mL/1
reported that 2 can be synthesized by heating a THF
solution of 1 under 40 atm of H2 for 48 h or by heating
1 in CH3OH under 6 atm of H2 for 6 h in a sealed tube.5b
Prolonged heating (40 h) of the solution at 80 °C led to
the formation of the metal carbonyl hydride complex
TpRu(PPh3)(CO)H (4) as the major product; a small
amount of 2 and a species which exhibited a 31P{1H}
singlet signal at δ 61.0 ppm were also present. The
carbonyl hydride complex 4 was identified by compari-
son of its 31P{1H} NMR signal and its hydride signal (δ
-11.83 ppm, J (HP) ) 26.8 Hz) with those of an
authentic sample.5b An independent high-pressure NMR
experiment showed that TpRu(PPh3)(H2)H in THF-d8
was slowly converted to 4 under CO2/H2 (8 atm/8 atm)
at 80 °C; at the conclusion of the experiment (20 h), GC-
MS analysis of gases in the headspace of the tube did
not show the presence of CO. It is probably true that,
in the reaction of 1 with CO2/H2 in anhydrous THF-d8,
the carbonyl hydride complex 4 was generated via the
intermediacy of TpRu(PPh3)(H2)H, although at this
stage we have no detailed mechanistic information for
this transformation. Separate studies, both experimen-
tal and theoretical, to unravel the mechanism of this
reaction are now in progress. In fact, there are prece-
dents in the formation of metal carbonyl from formic
acid or CO2/H2. For example, it was known more than
three decades ago that decarbonylation reactions of
formic acid by group III metal halide complexes could
be used to prepare carbonyl complexes.6 Wilkinson’s
complex Rh(PPh3)3Cl was converted to Rh(PPh3)2(CO)-
Cl when heated in polar aprotic solvents under CO2/H2
pressure; the presence of CO in the resulting gas
mixture suggested the occurrence of a reverse water-
gas shift reaction.7 It has recently been reported that a
pincer ligand iridium dihydride complex interacts with
CO2 to form an intermediate metallo formate complex,
which then undergoes disproportionation to give an
isolable metallo bicarbonate derivative and a carbonyl
hydride complex.8 The complex Cp*Ru(PCy3)Cl, which
failed to generate the coordinatively unsaturated ru-
thenium hydride species in its reaction with NaBH4,
was able to form the formato complex Cp*Ru(PCy3)(η2-
OCHO) when treated with carbon dioxide in the pres-
ence of NaBH4. More surprisingly, the initially formed
formato complex, in a sealed NMR tube in THF-d8, was
converted to a mixture of the metal carbonyl hydride
Cp*Ru(PCy3)(CO)H and formic acid in approximately
g
mL). f Solvent THF/H2O (17.5 mL/2.5 mL). The THF phase of
entry 7 was recycled. THF and H2O were added to make up a (15
mL/5 mL) THF/H2O solution, and 2 mL of NEt3 was also added.
0.5 mL of CH3CN added.
h
the hydrogenation of carbon dioxide to formic acid. We
report here the promoting effect of water in the 1-cat-
alyzed hydrogenation of CO2 and provide in situ high-
pressure NMR data and theoretical calculations to
explain this positive “water effect”.
Resu lts a n d Discu ssion
Hyd r ogen a tion of Ca r bon Dioxid e to F or m ic
Acid w ith 1. In the presence of triethylamine, and
under a total pressure of 50 atm (CO2/H2 ) 25 atm/25
atm), complex 1, in anhydrous THF, catalyzed the
hydrogenation of carbon dioxide to yield formic acid with
a low turnover number of 30 (entry 2, Table 1). Interest-
ingly, changing the solvent to THF/H2O led to great
enhancement of catalytic activity. Phase separation was
observed after the catalysis in these mixed-solvent
systems. In the one which contained 5% water (entry
3, Table 1), product distributions (as turnover numbers
(TON) in units of moles of product per mole of 1) in the
water and THF phases were 501 and 219, respectively.
Increasing the water content to 12.5% (entry 4, Table
1) significantly decreased the amount of product in the
THF phase (TON ) 687 in water phase and TON ) 48
in THF phase). When the percentage of water was
increased to 25%, the distribution of product in the
water phase was nearly exclusive, and only a trace
amount of the product was detected by 1H NMR
spectroscopy in the THF phase. On the other hand,
visual observation indicated that the catalyst had much
higher solubility in the organic phase than the aqueous
one, since the former was yellowish in color, and the
latter was colorless, before and after the catalysis. The
organic phase could be recycled and was found to be
active in the catalytic reaction, despite some loss of
activity (entry 8, Table 1)
In Situ High -P r essu r e NMR Stu d ies. To gain more
insight into the mechanisms of the catalytic CO2 hy-
drogenation reactions by 1 in anhydrous and hydrous
THF, we have carried out a series of in situ high-
pressure NMR studies. It is hoped that a greater
understanding of the enhancement effect of water in the
catalytic synthesis of formic acid from CO2 and H2 could
be acquired.
(6) Cleare, M. J .; Griffith, W. P. J . Chem. Soc. A 1969, 372.
(7) Koinuma, H.; Yoshida, Y.; Hirai, H. Chem. Lett. 1975, 1223.
(8) McLoughlin, M. A.; Keder, N. L.; Harrison, W. T. A.; Flesher, R.
J .; Mayer, H. A.; Kaska, W. C. Inorg. Chem. 1999, 38, 3223.
(a ) Rea ction of 1 w ith CO2/H2 in An h yd r ou s THF .
A THF-d8 (anhydrous) solution of 1, containing 25 equiv

1218 Organometallics, Vol. 20, No. 6, 2001
Yin et al.
a 1:2 ratio. Again, the detailed mechanism is not yet
clearly understood.9
8.42 ppm. 31P{1H} NMR monitoring over a 4 h period
at 100 °C showed the growing in of peaks at δ 68.0, 62.3,
61.8, 57.4, and 28.7 ppm. With the help of 1H NMR
spectroscopy, the first peak was assigned to TpRu-
(PPh3)(CO)H (4). The last peak was identified as the
signal of triphenylphosphine oxide, which was probably
formed by CO2 oxidation of PPh3. Unfortunately, we
have not been able to identify the other phosphorus
NMR peaks, which probably are due to some decompo-
sition products of the two initially formed complexes
(b) Rea ction of 1 w ith CO2 in An h yd r ou s THF .
As mentioned in the preceding paragraph, 1 reacts with
CO2/H2 in anhydrous THF-d8 to form TpRu(PPh3)(H2)H
(2) and a second species 3. To gain knowledge about the
structure of 3, we studied the reaction of 1 with
pressurized CO2 alone with NMR spectroscopy. It was
shown by 31P{1H} NMR spectroscopy that after an
anhydrous THF-d8 solution of 1 in a 5 mm Wilmad
pressure-valved NMR tube was subjected to 15 atm of
CO2 at room temperature for 19 h, complex 3 was
formed as the overwhelming product. The 1H NMR
spectrum of the 3 showed, in addition to the nine peaks
pertaining to the Tp ligand in the downfield region, a
sharp singlet at δ 8.17 ppm; this peak, integrated for
1H, could be ascribed to a formate proton. A sharp
singlet, which integrated for 3H and was due to coor-
dinated CH3CN, appeared at δ 2.25 ppm.
1
pertaining to the δ 54.8 and 57.8 ppm signals. The H
NMR spectrum also revealed the presence of a minute
quantity of TpRu(PPh3)(H2)H, but the amount was too
small to be detected by 31P{1H} NMR spectroscopy.
(d ) Rea ction of 1 w ith CO2 in Hyd r ou s THF . A
THF-d8 solution of 1 containing 10 vol % of water in a
5 mm Wilmad pressure-valved NMR tube was pressur-
ized with 10 atm of CO2; the 31P{1H} NMR spectrum of
the resulting solution recorded within 10 min showed
that 1 had already completely reacted. The spectrum
showed a singlet at δ 54.8 ppm, and a very small peak
The formato ligand in 3 was further evidenced by 13
C
NMR study of the reaction of 1 with 13CO2 (2 atm) at
room temperature in anhydrous THF-d8 for a prolonged
period of time (40 h). In the 13C{1H} NMR spectrum of
the resulting solution, the formato carbon of 3 appeared
as a singlet at δ 170.6 ppm. The off-resonance decoupled
spectrum showed a C-H coupling constant of 190.4 Hz.
When all the NMR data are taken together, 3 can be
identified as the formato complex TpRu(PPh3)(CH3CN)-
(η1-OCHO). In the reaction, a second but smaller for-
mate signal was detected in addition to that of 3; it
appeared as a singlet at δ 175.9 ppm in the 13C{1H}
NMR spectrum, and the signal was split into a doublet
1
at δ 57.8 ppm was also visible. The H NMR spectrum
of the same solution clearly showed the set of nine peaks
pertaining to the Tp ligand. A singlet at δ 2.30 ppm,
integrated for 3H, was assignable to the coordinated
CH3CN, and a singlet (1H) at δ 7.63 ppm could be
attributed to the hydrogen of a formate ligand. This last
peak disappeared when the deuteride complex TpRu-
(PPh3)(CH3CN)D (1-d ) was used in place of 1 in an
identical NMR study.
The 31P{1H} signal at δ 57.8 ppm gradually gained
intensity at the expense of the peak at δ 54.8 ppm. In
20 min time, the ratio of the intensities of the two peaks
was roughly 1:3. A concurrent 1H NMR spectrum
showed the free acetonitrile signal at δ 2.17 ppm; the
ratio of its intensity to that of the coordinated CH3CN
was also roughly 1:3. These two 31P{1H} signals were
identical with the two major peaks that first appeared
at the initial stage of the reaction of 1 with CO2/H2 (vide
supra). Therefore, it is clear that in the catalytic CO2
hydrogenation reactions in hydrous THF, complex 1
reacts rapidly with CO2 to produce 5 (corresponding to
the δ 54.8 ppm signal), and complex 6 (corresponding
to the δ 57.8 ppm signal) is then derived from 5 with
concomitant CH3CN dissociation from the latter. Un-
fortunately, it has not been possible to obtain a spec-
troscopically pure solution of 6, since 5 and 6 form an
equilibrium mixture.
1
with J (CH) ) 197.0 Hz in the H-coupled spectrum. We
attribute this smaller formate signal to the bidentate
formato ligand of TpRu(PPh3)(η2-OCHO).
Normal CO2 insertion into the metal-hydride bond
to generate a metal formate complex seems to be in
operation in most of the successful catalytic systems for
the CO2 hydrogenation reaction. Many previous studies
have shown that the stoichiometric reaction of carbon
dioxide with a metal hydride complex results in the
formation of a metal formate complex.10 Recently, a
number of theoretical studies describing this CO2 inser-
tion reaction have been reported.11
(c) Rea ction of 1 w ith CO2 a n d H2 in Hyd r ou s
THF . A THF-d8 solution of 1 containing 10 vol % of
water and 25 equiv of Et3N was pressurized with 25
atm each of CO2 and H2 in a 10 mm sapphire high-
pressure NMR tube. The 31P{1H} NMR spectrum taken
within 10 min at room temperature indicated that the
signal of 1 disappeared, and two major singlet signals
of roughly equal intensities were seen at δ 54.8 and 57.8
ppm. After heating at 100 °C for 10 min, the intensity
To verify the existence of the formato moieties in 5
and 6, reaction of 1 with 13CO2 in hydrous THF-d8 was
monitored by NMR spectroscopy. The 31P{1H} NMR
spectrum recorded after a THF-d8 (containing 5 vol %
of H2O) solution of 1 was subjected to 2 atm of 13CO2 at
room temperature for 1 h showed that 1 was completely
converted to a mixture of 5 and 6 in an approximately
2:1 ratio. The same solution exhibited, in the 13C{1H}
NMR spectrum, two singlets at δ 173.1 and 176.2 ppm,
which were assignable to the formato ligands of 5 and
6, respectively. Both peaks were split into doublets in
1
of the latter increased at the expense of the former. H
NMR spectroscopy now showed the formation of a small
amount of formic acid, as indicated by the singlet at δ
(9) Yi, C. S.; Liu, N. Organometallics 1995, 14, 2626.
(10) Some recent examples: (a) Albe´niz, M. J .; Esteruelas, M. A.;
Lledo´s, A.; Maseras, F.; On˜ate, E.; Oro, L. A.; Soler, E.; Zeier, B. J .
Chem. Soc., Dalton Trans. 1997, 181. (b) Whittlesey, M. K.; Perutz, R.
N.; Moore, M. H. Organometallics 1996, 15, 5166. Also see refs 8 and
9.
(11) (a) Hutschka, F.; Dedieu, A.; Eichberger, M.; Fornika, R.;
Leitner, W. J . Am. Chem. Soc. 1997, 119, 4432. (b) Musashi, Y.; Sakaki,
S. J . Chem. Soc., Dalton Trans. 1998, 577. (c) Musashi, Y.; Sakaki, S.
J . Am. Chem. Soc. 2000, 122, 3867.
1
the H-coupled 13C NMR spectrum; the former showed
an H-C coupling constant of 193.9 Hz, and that of the
1
latter was 198.2 Hz. The H{13C} NMR spectra of 5 and
6 showed the formate protons at δ 7.63 and 8.51 ppm,
respectively. In the off-resonance decoupled spectra, the

Ru-Catalyzed Hydrogenation of Carbon Dioxide
Organometallics, Vol. 20, No. 6, 2001 1219
Sch em e 1
Sch em e 2
proaching CO2 and a noncoordinating H2O molecule;
such an interaction increases the electrophilicity of the
carbon atom of the former and therefore facilitates its
insertion into Ru-H. Complex 5 then equilibrates with
6, and the two complexes are observable by high-
pressure NMR spectroscopy during the catalytic process.
In light of the Halpern axiom that observations of
“likely” intermediates in a catalytic cycle generally
signal, in fact, a nonproductive, sluggish loop of the
pathway,12 one would reasonably expect that the pro-
duction of formic acid via 5 and 6 is slow. Therefore,
the sequence in cycle A of Scheme 1 may not be the
productive route of the catalytic process. On the other
hand, we propose that the sequence in cycle B is
productive and is therefore the major route. The key
species in this cycle is 7, which is mainly formed by
displacement of the acetonitrile ligand of 1 by H2O. That
the availability of 7 has a profound effect on the overall
efficiency of the catalytic system is supported by the fact
that the presence of a small amount of CH3CN in the
system results in a very substantial decrease of the
turnover number of the catalytic reaction (entry 9, Table
1). A feature of this productive cycle is that a hydride
and a proton are simultaneously transferred to an
approaching CO2 molecule to form the formic acid in a
concerted manner via the intermediacy of 8. Noyori and
co-workers have proposed a simultaneous transfer of H-
and H+ from the Ru-amide bond to a carbonyl function
in the study of the ruthenium(II)-catalyzed hydrogen
transfer between alcohols and carbonyl compounds.13
Very recently, Morris et al. have reported similar
concerted dihydrogen transfer from cis hydride and
N-H groups of a ruthenium complex to the ketone or
imine substrate followed by heterolytic dihydrogen
splitting.14 Although it is not shown in cycle B of Scheme
1, elimination of formic acid from 8 probably generates
a transient hydroxo species, which then associates a
dihydrogen molecule and undergoes a σ-bond metathesis
reaction to regenerate 7 (Scheme 2).
Ch a r t 1
1
former was split into a doublet with J (HC) ) 193.7 Hz,
while the latter was split into a doublet with J (HC) )
1
198.5 Hz.
If the 31P, 13C, and H NMR data of 5 are taken into
1
consideration, it is obvious that the ligand set of the
complex consists of a Tp, a PPh3, a formate, and an
acetonitrile; therefore, the molecular formula that can
be inferred is TpRu(PPh3)(CH3CN)(η1-OCHO), which
seems unlikely because it is identical with that of 3. As
will be shown in a later section, theoretical calculations
show that the optimized structure of 5 is a hydrated
metal formate complex stabilized by hydrogen bonding
of the formate ligand with a water molecule (Scheme
1). Therefore, it seems reasonable that the molecular
formula TpRu(PPh3)(CH3CN)(η1-OCHO)‚H2O be as-
signed to 5.
The other formato complex 6, which is formed at the
expense of 5 and is in equilibrium with the latter, is
probably the aquo metal formate TpRu(PPh3)(H2O)(η1-
OCHO); this complex is also stabilized by a hydrogen-
bonding interaction between the formate ligand and
H2O, but intramolecularly, as opposed to the intermo-
lecular interaction in 5 (Scheme 1).
Mech a n ism s of CO2 Hyd r ogen a tion . The promot-
ing effect of water in CO2 hydrogenation has been
observed in a number of cases, including Pd-,2 Rh-,3 and
Ru-based4 catalytic systems. Some authors speculated
that water, as an ancillary ligand on the catalyst, could
accelerate the insertion of CO2 into M-H by hydrogen
bonding to one of the oxygen atoms of the incoming CO2
molecule3,4 (Chart 1).
In this section, we propose a mechanism aimed at
explaining the water effect in catalytic hydrogenation
of CO2 with 1, and theoretical data in support of the
proposed mechanism will be provided in the next
section.
σ-Bond metathesis reactions are well-recognized reac-
tions in the field of early-transition-metal, lanthanide,
and actinide chemistry;15 recent experimental16 and
theoretical17 findings, however, suggest that these reac-
tions could also occur with middle- and late-transition-
metal systems. A theoretical assessment on the σ-bond
In the CO2 hydrogenation reactions in hydrous THF,
complex 1 reacts readily with CO2 to form the hydrated
metal formate 5. That this insertion reaction is facile
is probably due to stabilization of the transition state
by a hydrogen-bonding interaction between the ap-
(12) (a) Halpern, J . Science 1982, 217, 401. (b) Halpern, J . Inorg.
Chim. Acta 1981, 50, 11.
(13) Yamakawa, M.; Ito, H.; Noyori, R. J . Am. Chem. Soc. 2000, 122,
1466.
(14) Abdur-Rashid, K.; Lough, A. J .; Morris, R. H. Organometallics
2000, 19, 2655.

1220 Organometallics, Vol. 20, No. 6, 2001
Yin et al.
metathesis reaction of water with a palladium hydride
complex to yield hydrogen has been reported; the
calculations show that such a reaction is feasible and
that it can in some instances be competitive with an
oxidative-addition/reductive-elimination sequence.18 Al-
though we have not been able to identify the intermedi-
ates 7 and 8, our theoretical calculations to be presented
in the next section strongly support the proposed
mechanism of the catalytic cycle.
In the case of CO2 hydrogenation in anhydrous THF,
the formato complex 3 is a stable intermediate detect-
able in in situ high-pressure NMR spectroscopy through-
out the catalytic process. Its formation is not as facile
as for its hydrated counterpart 5, probably due to the
relatively higher activation energy in the reaction
pathway. In view of the stability of 3, it is expected that
the generation of product from this species will not be
a low-energy process, and therefore, this catalysis,
unlike the one in hydrous THF in which there is another
low-energy pathway, has a low turnover number.
Th eor etica l Stu d y. Experimentally, it is found that
reactions of TpRu(PPh3)(CH3CN)H (1) with CO2 under
hydrous and anhydrous conditions differ significantly
in reaction rates. In hydrous THF solution 1 reacts with
CO2 rapidly to form the formato complex 5, while in the
anhydrous THF solution, the reaction goes more slowly
to generate complex 3. To understand the experimental
observations, theoretical calculations have been carried
out at the B3LYP level of density functional theory to
study the formation of 3 from the reaction of 1 with CO2.
(a ) In flu en ce of t h e Wa t er Molecu le on CO2
In ser tion in to th e Ru -H Bon d . Figure 1a shows the
calculated free energies for species involved in the
reaction under anhydrous conditions. The mechanism
here is similar to those found in the literature for CO2
insertion into M-H bonds.11 The transition state in-
volves the breaking of the metal-hydride bond and the
formation of both the C-H and Ru-O bonds. The CO2
molecule is reduced by the hydride through an electro-
philic reaction and a structural rearrangement to form
the η1-formato complex TpRu(PH3)(CH3CN)(η1-OCHO)
(3).
F igu r e 1. Free energy profiles for the reactions of TpRu-
(PH3)(CH3CN)(H) and CO2 (a) without and (b) with a water
molecule and for (c) transformation of TpRu(PH3)(OH)(H2)
to TpRu(PH3)(H2O)(H). The relative energies shown are in
units of kcal/mol.
structural feature of the transition state, we can infer
that the high reaction barrier is mainly determined by
the breaking of the metal-hydride bond (transferring
the hydrogen from the metal hydride to CO2) and the
stability of the product 3 is due to the formation of the
Ru-O bond. One thus expects that an enhancement of
the electrophilicity at the carbon center of CO2 can
reduce the reaction barrier and accelerate the metal
formate formation. Hydrous conditions are confirmed
to be able to serve the purpose.
To model the hydrous conditions, we study the same
reaction by adding a water molecule in our calculations.
Figure 1b shows the calculated free energies for the
reaction of 1 and CO2 in the presence of a water
molecule. The transition state (TS′) is now 26.2 kcal/
mol higher in free energy than the reaction 1′ + CO2.
Clearly, this activation energy is significantly lower
than that found for the reaction without a water
molecule (Figure 1a). Comparing the geometries of TS
and TS′ (see Figure 2), one sees that the Ru-H distance
in TS′ (2.361 Å) is longer than that in TS (2.322 Å),
but the C-H distance in TS′ (1.141 Å) is shorter than
that in TS (1.154 Å). The differences suggest that TS′
is an even later transition state regarding the hydrogen
transfer in this reaction. In addition, the Ru-O bonds
in TS′ and TS are the same (2.923 Å). Therefore,
interaction of the water molecule with CO2 enhances
the electrophilicity at the carbon center of the latter,
resulting in significant reduction of the reaction barrier.
The transition state (TS) of this reaction has been
optimized and confirmed by the frequency calculation.
The relative free energy of TS is 31.8 kcal/mol at room
temperature. Consistent with the experimental obser-
vation, this high activation energy implies a slow
reaction at room temperature. Structurally, TS is a late
transition state regarding the hydrogen-transferring
process. The Ru-H and H-C bonds in TS are 2.322 and
1.154 Å, respectively (see Figure 2). On the basis of the
(15) (a) Thompson, M. E.; Baxter, S. M.; Bulls, A. R.; Burger, B. J .;
Nolan, M. C.; Santarsiero, B. D.; Schaefer, W. P.; Bercaw, J . E. J . Am.
Chem. Soc. 1987, 109, 203. (b) Arndtsen, B. A.; Bergman, R. G.; Mobley,
T. A.; Peterson, T. H. Acc. Chem. Res. 1995, 28, 154. (c) Crabtree, R.
H. Chem. Rev. 1995, 95, 987. (d) Spessard, G. O.; Miessler, G. L.
Organometallic Chemistry; Prentice-Hall: Englewood Cliffs, NJ , 1996;
pp 369-373.
(16) (a) Hartwig, J . F.; Bhandari, S.; Rablen, P. R. J . Am. Chem.
Soc. 1994, 116, 1839. (b) Lee, J . C.; Peris, E.; Rheingold, A. L.; Crabtree,
R. H. J . Am. Chem. Soc. 1994, 116, 11014. (c) Schlaf, M.; Morris, R.
H. J . Chem. Soc., Chem. Commun. 1995, 625.
(17) (a) Musaev, D. G.; Mebel, A. M.; Morokuma, K. J . Am. Chem.
Soc. 1994, 116, 10693. (b) Siegbahn, P. E. M.; Crabtree, R. H. J . Am.
Chem. Soc. 1996, 118, 4442. (c) Hinderling, C.; Feichtinger, D.;
Plattner, D. A.; Chen, P. J . Am. Chem. Soc. 1997, 119, 10793.
(18) Milet, A.; Dedieu, A.; Kapteijin, G.; van Koten, G. Inorg. Chem.
1997, 36, 3223.
(b) Str u ctu r a l a n d En er getic Asp ects of Sp ecies
In volved in th e Ca ta lytic Cycle. To study the struc-
tural and energetic aspects of the species in Scheme 1,
we also calculated complexes 6-8, shown in Figure 2.
The calculations show that 6 + CH3CN is thermody-
namically more stable than 5 by 12.6 kcal/mol in free

Ru-Catalyzed Hydrogenation of Carbon Dioxide
Organometallics, Vol. 20, No. 6, 2001 1221
F igu r e 2. Optimized structures with selected structural parameters for species involved in the catalytic cycle shown in
Scheme 1.
energy, although experimentally they are found in
equilibrium. The inconsistency is believed to be due to
overestimation of the entropy contribution to the stabil-
ity of 6 + CH3CN because the electronic energy differ-
ence between 5 and 6 + CH3CN is 1.6 kcal/mol in favor
of the former. We suspect that the solvent effect might
also play a role here. However, the solvation energy
calculations based on the SCRF model (solvent )
acetonitrile) show that the solvent effect reduces the
energy difference only by 2 kcal/mol.
Structure 7 is calculated to be a hydride in which the
Ru-H bond is appreciably lengthened when compared
to that in 1. The lengthening can be related to its
dihydrogen bonding with a proton of the water ligand
(H- - -H ) 2.360 Å). Complex 8 shows a strong hydrogen
bond between one proton of the water ligand and one
of the two oxygen atoms of CO2. The free energy
calculations shows that complex 8 is 5.8 kcal/mol more
stable than 7 + CO2. Elimination of formic acid from 8
under the added Et3N/H2 to form the stable Et3N‚
HCOOH adduct regenerates complex 7. The reaction of
8 with H2 and NEt3 is unlikely to be a concerted
termolecular process because such steps are extremely
rare; it probably proceeds stepwise, and the sequence
of steps is not known. However, the actual barrier for
the regeneration of 7 from 8 should be smaller than the
assumed process, in which the formic acid is directly
removed from 8 in the presence of NEt3. The direct
removal of the formic acid from 8 to generate a transient
hydroxo species and the HCOOH‚NEt3 adduct requires
only 17.0 kcal/mol in free energy. This result indicates
that the process does not have a high barrier and should
be easily achieved.
According to Scheme 2, once the transient hydroxo
species TpRu(PPh3)(OH) is formed, the addition of H2
gives a dihydrogen hydroxo complex, which can be easily
transformed to complex 7 with a small free energy
barrier of 6.6 kcal/mol (Figure 1c).
Con clu sion
We have studied experimentally the promoting effect
of water in the catalytic hydrogenation of CO2 to formic
acid with TpRu(PPh3)(CH3CN)H. With the support of

1222 Organometallics, Vol. 20, No. 6, 2001
Yin et al.
ties of the formato proton of the product and the amido proton
of the internal standard were measured, and with help of a
calibration curve, the TON of the catalysis was derived. The
organic phase was analyzed in an analogous manner.
Com p u ta tion a l Deta ils. In our calculations, the PPh3
ligand was modeled using a PH3 group. Geometry optimiza-
tions and frequency calculations were carried out for all species
(reactants, precursors, products, and transition states) in-
volved in the reactions using the Gaussian 98 program19
installed on Pentium III personal computers with Linux (Red
Hat) operating systems.
Molecular geometries of the model complexes were opti-
mized at the Becke3LYP (B3LYP) level of density functional
theory.20 The LANL2DZ effective core potentials and basis
sets21 were used to describe Ru and P. Polarization functions
(ê(d) ) 0.340)22 were also added for P. Because of the large
size of systems studied, we used a STO-3G basis set for those
atoms which are not metal-coordinated in the Tp ligand and
for the hydrogen atoms in the PH3 ligands. The standard
6-31G** basis set was used for all other atoms. These basis
sets were designated as BS1. Since we were interested in the
relative energies, we expected that the errors produced by BS1
could be canceled when the relative energies were computed.
Indeed, single-point energy calculations using the better and
more consistent basis set described below indicated that the
changes in the relative energies were within 2 kcal/mol. The
better basis set (BS2) replaced STO-3G with 6-31G. With this
better basis set, single-point calculations based on the BS1-
optimized structures were performed to recalculate the ener-
gies (reported in the text). The entropy contributions were
evaluated through the BS1 frequency calculations of the
optimized structures.
a theoretical study, we have proposed a reaction mech-
anism to account for such an effect. One of the features
of the proposed catalytic cycle is the simultaneous
transfer of a hydride and a proton from the aquo
ruthenium hydride intermediate TpRu(PPh3)(H2O)H to
the CO2 molecule to give formic acid in a concerted
manner. The other feature is that the usual metal
formate intermediates resulting from CO2 insertion into
M-H in most of the successful catalytic CO2 hydrogena-
tion reactions with metal hydride complexes are prob-
ably not within the main catalytic cycle in the present
case. Failure to generate the aquo ruthenium hydride
intermediate in the absence of water is probably the
reason for the low activity of CO2 hydrogenation reac-
tions in anhydrous solutions.
Exp er im en ta l Section
Ruthenium trichloride, RuCl3‚3H2O, pyrazole, and sodium
borohydride were obtained from Aldrich. Triphenylphosphine
was purchased from Merck and was recrystallized from
ethanol before use. The complex TpRu(PPh3)(CH3CN)H was
synthesized according to published procedures.5b Solvents were
distilled under a dry nitrogen atmosphere with appropriate
drying agents (solvent/drying agent): tetrahydrofuran/Na-
benzophenone, diethyl ether/CaH2, acetonitrile/CaH2, n-hex-
ane/Na. High-purity hydrogen gas and carbon dioxide were
supplied by Hong Kong Oxygen.
Proton NMR spectra were obtained from a Bruker DPX 400
spectrometer. Chemical shifts were reported relative to re-
sidual protons of the deuterated solvents. 31P and 13C NMR
spectra were recorded on a Bruker DPX 400 spectrometer at
161.70 and 100.60 MHz, respectively. 31P chemical shifts were
externally referenced to 85% H3PO4 in D2O. 13C chemical shifts
were internally referenced to the residual peak of deuterated
solvents. High-pressure NMR studies were carried out either
in a commercial 5 mm Wilmad pressure-valved NMR tube or
a 10 mm sapphire HPNMR tube; the sapphire NMR tube was
purchased from Saphikon, Milford, NH, while the titanium
high-pressure valve was constructed at the ISSECC-CNR,
Firenze, Italy.
Ca ta lytic Hyd r ogen a tion of CO2 w ith Tp Ru (P P h 3)-
(CH3CN)H (1). The reactions were performed in a 100 mL
stainless steel autoclave. In a typical run, ∼0.013 mmol of 1
in 20 mL of THF/H2O (3/1 by volume) and 2 mL of NEt3 were
added to the autoclave. After being flushed with H2 three
times, the system was heated with stirring at 100 °C and under
50 atm of CO2/H2 (1/1). At the end of the required length of
time, the autoclave was cooled rapidly and vented carefully.
The resulting biphasic solutions were carefully poured into a
measuring cylinder, and the exact volumes of the two phases
were measured. A 0.2 mL aliquot of the aqueous solution was
syringed into a NMR tube, to which 0.3 mL of D2O and 6 µL
of DMF (internal standard) were added. The relative intensi-
Ack n ow led gm en t. We thank the Hong Kong Re-
search Grant Council (Project No. PolyU 32/96P) for
financial support.
OM000944X
(19) Frisch, M. J .; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.;
Robb, M. A.; Cheeseman, J . R.; Zakrzewski, V. G.; Montgomery, J . A.,
J r.; Stratmann, R. E.; Burant, J . C.; Dapprich, S.; Millam, J . M.;
Daniels, A. D.; Kudin, K. N.; Strain, M. C.; Farkas, O.; Tomasi, J .;
Barone, V.; Cossi, M.; Cammi, R.; Mennucci, B.; Pomelli, C.; Adamo,
C.; Clifford, S.; Ochterski, J .; Petersson, G. A.; Ayala, P. Y.; Cui, Q.;
Morokuma, K.; Malick, D. K.; Rabuck, A. D.; Raghavachari, K.;
Foresman, J . B.; Cioslowski, J .; Ortiz, J . V.; Stefanov, B. B.; Liu, G.;
Liashenko, A.; Piskorz, P.; Komaromi, I.; Gomperts, R.; Martin, R. L.;
Fox, D. J .; Keith, T.; Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.;
Gonzalez, C.; Challacombe, M.; Gill, P. M. W.; J ohnson, B. G.; Chen,
W.; Wong, M. W.; Andres, J . L.; Head-Gordon, M.; Replogle, E. S.;
Pople, J . A. Gaussian 98, revision A.7; Gaussian, Inc.: Pittsburgh, PA,
1998.
(20) (a) Becke, A. D. J . Chem. Phys. 1993, 98, 5648. (b) Miehlich,
B.; Savin, A.; Stoll, H.; Preuss, H. Chem. Phys. Lett. 1989, 157, 200.
(c) Lee, C.; Yang, W.; Parr, G. Phys. Rev. B 1988, 37, 785.
(21) Hay, P. J .; Wadt, W. R. J . Chem. Phys. 1985, 82, 299.
(22) Huzinaga, S. Gaussian Basis Sets for Molecular Calculations;
Elsevier: Amsterdam, 1984.